HOW TO MEASURE NO EFFECT. PART I: TOWARDS A NEW MEASURE OF CHRONIC TOXICITY IN ECOTOXICOLOGY. INTRODUCTION AND WORKSHOP RESULTS

Abstract

The NOEC (no observed effect concentration) is widely used as summary statistic of ecotoxicity tests. In setting standards for risk evaluation, it is normally applied as if it were an estimate of the NEC (no effect concentration). Recently, a workshop was organized in the Netherlands to evaluate the merits of the NOEC and those of potential alternatives in ecotoxicity testing. A short introduction on the NOEC and a survey of the results of that workshop are given. Two parametric alternatives for the NOEC were discussed: (i) the ECx, i.e. the estimated concentration causing an effect of x per cent compared to the control; (ii) the parNEC, i.e. the estimate of the NEC in a parametric threshold model of the concentration–response relationship. No clear-cut preference for either of these two alternatives emerged. The ECx was considered attractive because this parameter is expected to be less model dependent than the parNEC approach and the method is well established. The parNEC was found attractive because it embraces the concept of no effect. The main conclusions of the workshop were: (i) NOEC values should always be accompanied by the minimum observable difference under the given test conditions. The value of the observed parameter in the NOEC compared to the control should also be given. (ii) The NOEC should (in due time) be replaced by another measure for (almost) no effect. (iii) Research is needed to choose between the two potential alternatives for the NOEC: an ECx value and a parametric NEC estimate. The following two areas of research were recommended: a re-evaluation of old ecotoxicity databases, involving the determining of the NOEC, EC5, EC10 and NEC estimate for these data sets and comparing their values; evaluation of several ECx and NEC models with the aim of determining the accuracy and the model dependence of these estimates using a large set of computer simulated data. (iv) If an ECx value is chosen, the preferred vaue of x would be 5 or 10 per cent. © 1997 John Wiley & Sons, Ltd.